Meta has defeated a lawsuit that alleged it breached the copyright law protecting authors by using their books without permission to train its AI large language model, Llama.
It is the second legal victory for AI tech companies in a week after a US court ruled Anthropic had not breached copyright by training its artificial intelligence large language model Claude on authors’ work.
In the Meta ruling, San Francisco US district judge Vince Chhabria said that the authors, which included Sarah Silverman and Ta-Nehisi Coates, had not proven the company’s use of AI would cause “market dilution” by flooding the market similar work, which meant that Meta’s use of their literature was considered “fair use” and not subject to copyright law.
“This ruling does not stand for the proposition that Meta’s use of copyrighted materials to train its language models is lawful,” Chhabria said. “It stands only for the proposition that these plaintiffs made the wrong arguments and failed to develop a record in support of the right one.”
The judge dismissed as “nonsense” Meta’s claim that the public interest would be “badly disserved” if tech companies were not protected from scraping copyrighted material to train large language models.
This opens the door to future claims that can better illustrate how LLM training harmed the author’s work.
In the Anthropic case, another San Francisco district judge said Anthropic made “fair use” of books by the writers Andrea Bartz, Charles Graeber and Kirk Wallace Johnson to train Claude.
Judge William Alsup said that Anthropic model’s use of books was akin to a “reader aspiring to be a writer” and not necessarily replace or supplant authors.
However, Judge Alsup warned that copying and storing more than 7 million pirated books in a central library has infringed copyright laws and was nor “fair use”.
Authors, media companies and other creators have brought several lawsuits before US courts testing the boundaries of copyright law and whether their work can be used without permission.
